Reading raw data

Raw poses are published directly with valid device frames; the IMU is published directly when the device frame contains the corresponding data. They do not need retargeting scheduling, and are not generated or rate-limited by run_forever(), run_background(), or update().

Glove pose

Read GlovePose through device.pose():

• meta: source, time, optional sequence, and frame ID.
• payload.root_pose: root node position and XYZW quaternion.
• payload.fingertip_world and fingertip_local: world/local poses of the five fingers.
• raw_valid_mask, held_mask, and quality_flags: quality information.

For the complete fields, see GlovePose and GloveFeaturePayload.

IMU (experimental)

Read the raw IMU counts provided by the glove through device.imu():

auto device = glove.device();
auto imu = device.imu();
auto imu_subscription = imu.subscribe(
    [](const qnbot::Sample<qnbot::GloveImu>& sample) {
        const auto& value = sample.value.payload;
        std::cout << value.valid << ' ' << value.gyroscope_raw[0] << ' '
                  << value.accelerometer_raw[0] << '\n';
    });

bool started = false;
try {
    device.start();
    started = true;
    run_application();
    device.stop();
    started = false;
    glove.close();
} catch (...) {
    if (started) {
        device.stop();
    }
    glove.close();
    throw;
}

gyroscope_raw and accelerometer_raw are each three std::int16_t values, corresponding to the three axes in order. The SDK performs no unit conversion, calibration, or pose fusion. When valid == false, the raw counts are still retained and published. The IMU is currently an experimental feature; the data format, supported devices, and public behavior may change later. For the complete fields, see GloveImu and GloveImuPayload.

Device status

device.status() is also produced directly by the device rather than computed: whether it is connected, whether data is stale, the last error and the reconnect count. See Diagnostics & errors for the fields and how to use them for diagnosis.